Ajay Bagga flags risks as hyperscalers pour over $1 trillion into AI infrastructure

As global technology companies race to scale artificial intelligence infrastructure, market expert Ajay Bagga has flagged deep structural shifts that could reshape the tech sector — and unsettle investors along the way. 

In a series of posts on X (formerly Twitter), Bagga warned that the explosive capital expenditure by hyperscalers is fundamentally altering the nature of large software firms, transforming them into capital-intensive hardware operators with limited financial visibility. 

Software firms enter a capital-heavy era 

Bagga pointed out that hyperscalers are expected to spend $450 billion in 2025, with plans to deploy another $600 billion in 2026, largely to expand AI-driven infrastructure. This unprecedented spending spree, he argued, blurs the traditional distinction between software and hardware businesses. 

“All the big software companies are hardware companies now — capital-intensive,” Bagga wrote, cautioning that markets lack clarity on recurring and replacement maintenance capital expenditure. He added that uncertainty over revenue visibility, faster technology obsolescence, commoditised pricing, and evolving depreciation standards could weigh heavily on valuations. 

According to Bagga, the accelerated pace of innovation in AI hardware raises fresh concerns around shorter asset lifecycles, forcing companies to continuously reinvest just to stay competitive. 

Critical bottleneck 

Beyond chips and servers, Bagga highlighted power generation, transmission, and storage as critical enablers of the AI and data centre rollout. Hyperscalers, he noted, are increasingly scrambling to lock in long-term electricity supply deals to meet soaring energy demands. 

Citing recent developments, Bagga said Meta has struck multiple agreements with nuclear power companies to support its AI ambitions. These include a 20-year power purchase deal with Vistra, as well as commitments with Oklo and TerraPower, which are developing small modular nuclear reactors expected to come online in the early 2030s. 

The moves underscore how energy security is becoming central to AI scalability, pushing technology firms into long-term infrastructure partnerships once considered outside their core business. 

AI, Algorithms & transparency question 

Bagga also echoed broader concerns about transparency in the AI era, referencing comments by investor Michael Burry. He warned that societies are increasingly guided by algorithms that even their creators may not fully understand. 

“In the AI era, transparency isn’t just a ‘nice to have’—it’s a survival requirement for a stable society,” Bagga quoted, highlighting the governance and accountability challenges emerging alongside rapid technological adoption.